Research Institute "Quantum"

Safe Reinforcement Learning focuses on developing optimal policies while ensuring safety. A popular method to address such task is shielding, in which a correct-by-construction safety component is synthetised from logical specifications. Recently, shield synthesis has been extended to infinite-state domains, such as continuous environments. This makes shielding more applicable to realistic scenarios. However, often shields might be unrealizable because the specification is inconsistent. In order to address this gap, we present a method to obtain simple unconditional and conditional explanations that witness unrealizability, which goes by temporal formula unrolling: bounded strategy search. In this paper, we show different variants of the technique as well as its applicability

Abstract Let p be a prime number, p ≥ 3. We consider the set of decompositions of a p-logic function into a product of functions with disjoint subsets of variables obtained by means of linear substitutions of arguments. Each decomposition of this kind is associated with a decomposition of the vector space into a direct sum of subspaces. We present conditions under which such space decomposition is unique up to rearrangement of subspaces. Also, a criterion for such product to be balanced is given.

A search for an optimum technique for studying the electrical characteristics of thin n/p-InxGa1–xAs semiconductor layers with different doping levels has been carried out. The primary task has been to measure the main electrical characteristics by different methods using resistivity (conductivity), majority carrier concentration, dependence of the main electrical parameters on the doping type and level, and their comparison. Using the example of the p- and n-In0.01Ga0.99As solid solutions grown by MOCVD, a technique for studying the main electrical characteristics of the epitaxial layers has been proposed, which takes into account the estimated homogeneity on large-area samples. Results obtained by different methods, including photoluminescence, contactless surface resistivity measurement, van der Pauw (Hall effect), electrochemical capacitance–voltage profiling, and in situ control, have been compared. Basing on the results obtained and comparison with the literature data, conclusions have been drawn concerning the need, sufficiency, and complementarity of the methods for controlling and studying semiconductor epitaxial structures.

The Khrapchenko method of finding a lower bound for the complexity of binary formulas is extended to formulas in $$k$$ -ary bases. The resulting extension makes it possible to evaluate the complexity of linear Boolean functions and a majority function of $$n$$ variables when realized by formulas in the basis of all $$k$$ -ary monotone functions and negation as $$\Omega(n^{g(k)})$$ , where $$g (k)=1+\Theta(1/\ln k)$$ . For a linear function, the complexity bound in this form is unimprovable. For $$k=3$$ , the sharper lower bound $$\Omega(n^{1.53})$$ is proved.

Solar panels are the primary power supply source for the onboard equipment of the most modern and the most promising spacecraft. The InGaP/InGaAs/Ge triple-junction solar cells (SCs) are the photogenerating part of solar panels for space application. Such SCs convert efficiently the solar radiation spectrum into electrical energy due to photon absorption in semiconductor layers with different widths of the bandgaps. In addition, these ones have great radiation resistance. However, the high cost of triple-junction solar cells is one of the limiting criteria for the solar panels developers when choosing them, despite all the advantages. Consequently the reduction of technological losses requires great attention in the manufacture of solar cells. The aim of experimental annealing was optimization of the SCs electrical characteristics and reduction of technological losses. The influence of experimental annealing was estimated using the I-V characteristic, external quantum efficiency and electroluminescence measurements. Isochoric annealing made it possible to optimization of SC power by 1,26%, in comparison with the other types of annealing used in this work.

One of the existing methods for calculating the total resistance of a solar cell (SC) model with series resistance applied to SC cascades was considered. A balance of optical and ohmic losses was found, the optimal values of the width of the strips of the contact grid and the distance between them were determined by comparing the experimental data obtained on the basis of the dependences of shadowing by the contact grid and the resistance of space SC based on InGaP/InGaAs/Ge heterostructures with an area of 27,5 cm2 geometric parameters of the contact grid. Accordingly, the values were from 15 µm to 19 µm and from 750 µm to 1000 µm with a shading of 1.82 %−1.93 %. The above comparisons with similar parameters of SC of world manufacturers allow us to talk about the possibility of using this technique for SC of other sizes.

The authors devoted the paper to machine learning methods and algorithms for the supercomputer job execution time predicting. The supercomputer systems statistics for multiple-access shows that the actual execution time for most jobs substantially diverges from the time requested by the user. This reduces the scheduling job efficiency because inaccurate estimates of execution time lead to a suboptimal schedule for job launches.

В работе предпринимается обзор современного состояния теории быстрых алгоритмов умножения чисел и многочленов. Рассматривается процесс эволюции методов умножения от первых блочных алгоритмов Карацубы и Тоома 1960-х гг. к методам 1970-х гг., опирающимся на дискретное преобразование Фурье (ДПФ), и далее к новейшим методам, разработанным в 2007–2019 гг. Современные методы умножения сочетают использование специальных алгебраических структур, переход к приближенным вычислениям, особые формы преобразований Фурье: многомерное ДПФ, аддитивный аналог ДПФ. Эти и другие существенные для быстрых методов умножения концепции подробно рассматриваются в настоящем обзоре. Отдельно предусмотрено введение в теорию ДПФ с извлечением необходимых для изложения материала фактов. В заключительной части обзора приводятся краткие сведения о результатах в области параллельных алгоритмов умножения, аккуратных оценок сложности базовых методов умножения, алгоритмов умножения в реальном времени, мультипликативной сложности умножения многочленов над конечными полями. Отмечены модели вычислений, в которых умножение имеет линейную или квадратичную сложность.

Aim. The article deals with studying the changes in binocular pupillometry to establish vegetative activity in athletes after high exercise stress. Materials and methods. 25 male athletes aged 19–34 years with no health limitations participated in the study. We estimated a set of the commonly accepted parameters of pupillometry (pupil diameter, pupil reaction initiation time, pupil constriction diameter, pupil constriction velocity) by measuring pupil reaction to a light stimulus with the help of a non-contact pupilometer (KVAZAR, Moscow). The examination was conducted prior to physical stress and 5, 15, and 30 minutes after. Intertrack-750 treadmill (SchillerAG, Switzerland) was used for testing. Results. The results obtained in a pupillometric study prove that there is a predominance of the sympathetic component in athletes at maximal load accompanied by emotional stress. After physical stress accompanied by fatigue and nervous system exhaustion, there is a pronounced predominance of the parasympathetic component. Conclusion. The data obtained prove that binocular pupillometry can be effectively used in sports medicine to estimate vegetative tone and regulatory mechanisms in athletes.

A comparative analysis of the formation and decomposition processes and the component composition of the phases of films that passivate a nickel electrode during its dc polarization in sulfuric acid solutions with a varied value and polarity of the applied potential is performed.